Dynamical relaxed directional method for fuzzy reliability analysis. (December 2021)

• Record Type: Journal Article
• Authors: Bagheri, Mansour; Hosseini, Seyed Abbas; Keshtegar, Behrooz
• Is Part Of: Structures. Volume 34(2021)
• Page Start: 169
• Page End: 179
• Publication Date: 2021-12
• Subjects: Epistemic uncertainty -- Fuzzy reliability analysis -- Dynamical relaxed directional method -- Probabilistic method -- Cuckoo search algorithm; Structural engineering -- Periodicals; 624.1
• DOI: 10.1016/j.istruc.2021.07.043
• Languages: English
• ISSNs: 2352-0124

Abstract

Highlights: A novel bi-loop fuzzy reliability method is presented for engineering structures under epistemic uncertainties. A robust iterative formulation is proposed for evaluating the reliability index using dynamical relaxed factor. The improved cuckoo search method is utilized to verify the uncertainty interval of reliability index. The proposed method is robustly preformed lowest computational burden than existing iterative reliability method. Abstract: As an efficient and robust tool in handling uncertainties, the first-order reliability method (FORM) is proposed for engineering problems under epistemic uncertainty. The method, namely dynamical relaxed directional method (DRD) coupled with an improved cuckoo search algorithm (ICOA) as DRD-ICOA consists of two numerical local and global analyzers, in which local analyzer performs reliability estimation to appraise the reliability index using the DRD approach. Whiles the global analyzer carries out the epistemic uncertainty of random variables using the alpha level set optimization by applying an ICOA. The DRD is formulated using the relaxed approach with a descent search direction by the dynamical sensitivity vector. The dynamical sensitivity vector is formulated using the sufficient decent conditions to guarantee the stability of FORM formula in structural fuzzy reliability analysis. Applicability of proposed numerical method of DRD-ICOA was examined via four engineering examples and was compared with several reliability methods-based FORM methods. Investigation results showed that the DRD Form could robustly converge to exact solutions and deal with real engineering problems via low computational burden.